With the advent of wideband transmitters, receivers, and dual-polarization, monopulse antennas, it is possible to image radar targets in several dimensions simultaneously. Wideband, frequency agility radar processing makes it possible to image a point scatterer in the range dimension. Coupling relative target rotation pnd frequency agility, it is possible to image a point scatterer in two orthogonal dimensions (range, cross-range). Dual channel monopulse processing coupled with wideband frequency agility makes it possible to measure the spatial position of an isolated point scatterer (range, azimuth, elevation). The complete electromagnetic characterization of the target (at an aspect angle) is accomplished by measuring the target's four polarization responses in a polarization basis. Measuring all target aspects is accomplished by continuous turntable measurements. Uniting frequency agility, polarization agility, monopulse processing, and angular processing, the target may be imaged in one, two, and three spatial dimensions simultaneously for each of the polarization pairs. One-dimensional imaging is called high-range resolution (HRR) processing; two-dimensional imaging is either Synthetic Aperture Radar (SAR) or Inverse Synthetic Aperture Radar (ISAR); three-dimensional imaging is monopulse imaging.

Signal processing for image reconstruction in synthetic aperture radar (SAR) historically has been based on Fourier transform techniques. One reason for this is the fact that, at the time when SAR was invented in the early 1950's and for some time after that, the only way to process the huge amounts of data in a reasonably expeditious manner was to use optical techniques, such processors being based, at least in part, on Fourier optical principles. With the advent of digital processing in recent years, the existence of efficient algorithms for the computation of the discrete Fourier transform has continued to offer compelling reasons to use Fourier-type inversion methods. Additionally, geometrically-related simplifications in most analyses have engendered the assumption of plane waves being present over the ground patch being imaged, again encouraging the use of Fourier techniques. In applications where the distance of the radar from the ground patch is very large compared to the size of the ground patch to be imaged, such processing is appropriate, though still an approximation. In other cases, the wavefront curvature cannot be ignored, and other steps must be taken in order to yield high-quality imagery. Recent investigations into the use of convolution-backprojection algorithms modified from computer-aided tomography have proved fruitful in correcting for wavefront curvature in monostatic SAR. This paper reports on similar success in bistatic SAR. There appear to be other applications that could benefit from other adaptations of the convolution-backprojection idea.

The chirp z-transform was proposed in for replacing the conventional polar format interpolation used in synthetic aperture radar (SAR) signal processors. However, the proposed method suffers from important limitations. The purpose of this conference paper is to describe the problems and limitations with the methods described in. A brief review of conventional approaches is included to provide a basis for comparison.

A new algorithm has been proposed for spotlight-mode synthetic aperture radar in a recent correspondence by W. Lawton. This new processing scheme assumes that Fourier transform samples of the obtained projections lie on a trapezoidal grid. The spatial domain image is produced through a series of convolutions and DFT's, all performed using FFII's. The geometry of the trapezoidal grid makes this fast algorithm possible, which requires 0 (N²logN) multiplications for an NxN image. In this paper we show that the Lawton algorithm implements a form of trapezoidal-to-Cartesian interpolation followed by a 2-D FFT. We then derive a closed-form approximate expression for the point-target response of the algorithm with Hamming weighting of the acquired data. Next, we simulate reconstructions of point targets and of extended targets composed of groups of point targets. The effects of windowing and incorporation of a Jacobian weighting factor are deter-mined. We also consider the effect on image quality of moderate levels of sampling jitter in the A/D and of deviations of actual data collection angles from those intended. These errors serve to provide Fourier samples of the reflectivity on a grid that deviates from the assumed trapezoidal raster. Overall, the Lawton algorithm is found to be robust, and to produce high quality imagery.

The polar format approach to SAR image formation requires data to be interpolated from a warped grid onto a cartesian lattice. In general, this requires that data be interpolated between varying sampling rates. In this paper we define and justify frequency-domain optimaliy criteria for polar format interpolators and describe an approach to designing the corresponding k digital filters.

It has been suggested that synthetic aperture radar (SAR) images obtained from platforms such as SEASAT are subject to potential degradation by ionospheric-induced phase errors. This premise is based upon data from various satellite experiments that indicate large levels of phase scintillation in auroral zone data. Current models for phase errors induced by the ionosphere suggest that the phase error power spectrum is power law. This implies that the resulting phase errors contain significant components up to the Nyquist limit. Traditional sub-aperture based autofocus techniques, designed to correct uncompensated platform motion errors, are inadequate due to their inability to estimate higher order error terms. A new non-parametric phase error correction scheme developed at Sandia National Laboratories, however, has been demonstrated to remove phase errors of arbitrary structure. Consequently, our new algorithm is a viable candidate for correcting ionospheric phase errors. In this paper we show examples of SAR images degraded by simulated ionospheric phase errors. These images demonstrate that such errors cause smearing with complicated sidelobe structure. Restoration of these images via the new algorithm illustrates its superiority to classical sub-aperture based autofocus techniques.

Two techniques for using distributed clutter for calibrating a polarimetric radar are given: one uses clutter only but requires knowlege of theoretical clutter polarization ratios; the other requires the use of an in-scene trihedral but requires no foreknowledge of clutter statistics. These techniques are developed and contrasted with the more traditional technique which uses a set of in-scene calibration reflectors, such as trihedrals and dihedrals. A method is proposed and evaluated for estimating and removing the effects of additive noise on the measured clutter covariance matrix. Results are given of the application of this technique to MIT Lincoln Laboratory's Advanced Dectection Technology Sensor (ADTS) synthetic aperture radar.

The most popular optimality criteria for digital filters is minimax (Chebyshev) in each band. In particular, the Parks-McClellan algorithm is very popular for FIR digital filter design. However, signal processing operations for synthetic aperture radar (SAR) require other types of filters. The design of FIR digital filters for SAR image formation applications is presented in this paper.

Conventional microwave and futuristic optical radar imaging requires estimating a function from limited Fourier information and a priori constraints. Linear problems include estimation from complex Fourier samples that are irregularly spaced on a polar grid (spotlight SAR), on a sinusoidal grid (laser SAR), or on a random grid (anti-aliasing phased array). Nonlinear problems include the famous phase retrieval problem of estimating a two-dimensional band limited function from its approximate Fourier modulus over a spatial region. Several deterministic and iterative methods for linear problems are described and their algorithmic complexity is discussed. A new method for attacking the infamous 'stagnation' problem that characterizes iterative methods for the phase retrieval problem is described. The method, which originated from consideration of spin-glass models of statistical lattice physics, incorporates partial information about the zero-set of the analytic continuation of the squared Fourier transform modulus in order to break the 'twin-object' symmetry responsible for stagnation.

Speckle is a major cause of degradation in synthetic aperture radar (SAR) imagery. With the availability of fully polarimetric SAR data, it is possible to use the three complex elements (HH, HV, VV) of the polarimetric scattering matrix to reduce speckle. This paper derives the optimal method for combining the elements of the scattering matrix to minimize image speckle; the solution is shown to be a polarimetric whitening filter (PWF). A simulation of spatially correlated, K-distributed, fully polarimetric clutter is then used to compare the PWF with other, suboptimal speckle-reduction methods. Target detection performance of the PWF, span, and single-channel |HH| detectors is compared with the optimal polarimetric detector (OPD). Finally, a new, constant false alarm rate (CFAR) detector (the adaptive PWF) is proposed as a simple alternative to the OPD for detecting targets in clutter. This algorithm estimates the polarization covariance of the clutter, uses this covariance to construct the minimum speckle image, and then tests for the presence of a target. An exact theoretical analysis of the adaptive PWF is presented; the algorithm is shown to have detection performance comparable with that of the OPD.

There has been much work ih Vibe area of data compression since the fundamental source coding theories were defined by Shannon. ' Most of the work has been done on speech with a growing percent on optical imagery. Only recently has work on Synthetic Aperture Radar (SAR) data compression begun to mature. With the high data rate of a typical SAR system, methods for compression have become important. In this paper, we focus on the source coding technique of Vector Quantization (VQ) for use in a SAR imaging environment. Several types of vector quantizers are presented and discussed. It is shown that VQs can achieve a minimum of 2:1 compression on SAR phase history data without any loss of quality.

An algorithm to detect the presence of ground-moving targets in synthetic aperture radar (SAR) images has been investigated. Its performance is independent of the direction of the target velocity. Called the notch algorithm for its method of processing, it makes use of the shadow characteristics of a target moving over a clutter background in a unique way. While the image of a moving target may be displaced from its true position in a SAR image, the shadow cast by that target will not be shifted, although it will be de-focused and difficult to detect with conventional SAR processing. The notch algorithm attempts to find the target's ground-track and estimate its velocity vector using a single, coherent aperture of radar data. This information is critical to finding the moving target in the image of the scene and to indicating phase compensations required to begin to re-focus the target. Ground-track information can be obtained by applying the notch algorithm to the same data set several times while varying the algorithm parameters, followed by a comparison of the resulting images with the conventionally processed image. In this paper we describe the basis for the algorithm and present simulation results showing the performance over a range of target velocities. We also suggest several methods to improve detection.

The detection of moving targets with Synthetic Aperture Radars (SAR) is usually carried out by means of methods based on a Doppler filtering. The targets are detected if their Doppler frequency spectrum falls outside of the clutter spectrum, where the clutter consists, in this case, on the returns from the ground. These methods however present some drawbacks, namely: a) targets with low radial velocity, with respect to the radar, are not detected and, b) in order to create a certain visibility region in the Doppler frequency domain, the Pulse Repetition Frequency (PRF) must be taken sensibly higher than the clutter bandwidth and this entails a reduction of the monitorable swath and then a loss on the possible radar coverage. In order to overcome the afore-mentioned shortcomings, a technique for detecting moving targets whose spectrum is embedded in the clutter spectrum is presented in this paper. The technique is based on the difference in the Doppler frequency rate and spatial correlation between moving and fixed targets. The received signal is first filtered in an adaptive way for improving the power ratio between moving and fixed target. The adaptive filter is then followed by a Doppler-rate filters' bank, each filter being matched to a particular Doppler-rate. The performances of the proposed technique are assessed by means of a simulation program.

This paper describes a new gradient algorithm for enhancing linear features in synthetic aperture radar imagery. The algorithm is based on finding the maximum average gradient over a local window centered on a pixel. The algorithm also outputs the local direction of linear features. The algorithm attempts to compensate for decorrelation resulting from the speckle in radar imagery by considering larger pixel neighborhoods or processing windows.

Vision Guidance Update (VGU) is a Synthetic Aperture Radar (SAR) multi-image exploitation technique which uses radar vision derived from range-angle imagery of stationary ground features to provide precision platform position and velocity estimates. The concept addresses applications to both navigational update and guidance functions for systems using SAR as an autonomous, long-range, standoff, all-weather, day/night sensing capability. The changes in perspectives of ground features detected in successive SAR spotlight images are analyzed and the coupled geometry problem solved to determine precision platform velocity and position estimates relative to any imaged point on the ground. A SAR VGU theory and algorithm were developed that utilizes two features in two images for deriving the platform kinematic estimates. Realistic SAR finite resolution effects lead to estimation errors. Errors in the knowledge of platform position and velocity, generated from navigation system errors, produce imaging geometry and image formation errors. Tolerance to reasonable platform kinematic errors is provided by iterating the SAR VGU algorithm to improve and provide precision estimates of all geometry, system, and kinematic parameters. The VGU theory and algorithm are presented with the initial modeling assumptions. Initial simulation results are qualitatively summarized.

This paper investigates the use of synthetic discriminant functions (SDFs) for pattern recognition applicable to computer generated synthetic aperture radar (SAR) imagery. Complex valued and real valued projection SDFs are constructed and tested to determine their effectiveness as SAR pattern recognizers. In addition, SDFs mapped to a position, scale, rotation invariant (PSRI) feature space are tested and compared to conventional projection SDFs.

The millimeter wave (MMW) frequency band offers a potential solution to the problem of identifying, tracking, and engaging targets in adverse weather conditions. Seekers employing MMW radar require small, lightweight designs using gimballed RF electronics and an RF subsystem consisting of a waveform generator, a transmitter, and a multichannel receiver.

The Advanced Technology Millimeter Wave Seeker Testbed (ATMMWST) may be characterized by these descriptors: 1) High range resolution (HRR) via synthetic, coherent processing 2) Complete polarization scattering matrix in a circular basis 3) Dual-plane sum-and-difference monopulse with complex angle processing. This seeker technology is coupled with statistical pattern recognition algorithms for target/clutter discrimination and tracking algorithms for guidance signal generation. The algorithms are embedded in the signal processing software/hardware system. The ATMMWST system consists of a seeker, a signal processor, an instrumentation system and data recording system, and an independent line-of-sight reference system (ILOSRS). The system is used in both tower and captive flight programs to collect target signatures and to demonstrate various aspects of the mission scenario.

This paper describes the development and preliminary testing of an imaging passive microwave radiometer operating in the 10 to 85 GHz range specifically for precipitation retrieval and mesoscale storm system studies from a high altitude ER-2 aircraft. The instument is referred to as the Advanced Microwave Precipitation Radiometer (AMPR). The primary goal of AMPR is the exploitation of the scattering signal of precipitation at frequencies near 10, 19, 37, and 85 GHz to unambiguously retrieve precipitation and storm structure intensity information in support of space sensors, as well as, storm-related field experiments onboard the ER-2 in 1989/90. A unique feature of the AMPR instrument is the multi frequency feedhorn (identical to the SSM/I space instrument) used at 19.35, 37.0, and 85.5 GHz. The feedhom provides dual polarization ports at each of these three frequencies. The 10.7 GHz channel uses a separate horn/lens antenna with single polarization only. The AMPR scanner is designed to image over an angular range of ±40° about the nadir position. The AMPR will be used onboard the ER-2 to perform underflights of the SSM/I space instrument with enough spatial coverage to make meaningful comparisons of the data. The AMPR will be the only aircraft sensor with the capability to image at the same frequencies as the SSM/I and at an altitude high enough for precipitation studies over land.

The National Aeronautics and Space Administration, (NASA) in conjunction with other international civilian space agencies (ESA, NASDA), has embarked on an ambitious plan to deploy a number of large earth orbiting platforms. The space station to be launched in 1993 will be in an equitorial orbit, manned by an astronaut crew and carrying a set of experiments that require operator interaction, servicing, or will only operate for a short period of time (< 1 year). As a complement to the space station, a series of unmanned polar orbiting platforms (POPs) are also under development (Table 1). These platforms will carry a suite of remote sensing instruments designed to observe characteristics of the Earth's surface, its atmosphere, or for planetary and deep space observation.

Simulated signatures of generic airborne targets are presented in this paper. In particular, two-dimensional images of target shapes as a function of frequency and bistatic angle (between the transmitter and receiver lines-of-sight) are investigated. The theoretical model is based on the classical theories of physical optics and diffraction. Coherent radar cross section is first simulated for monochromatic waves. Image response is then obtained by Fourier transforms infrequency and time. Accurate bistatic radar cross section (RCS) is simulated by evaluating both the physical optics and the diffraction components with a uniform asymptotic analysis.

Synthetic Aperture Radar (SAR) images of ship wakes have, under calm wind conditions, contained a "narrow vee" artifact of included angle small relative to that of the Kelvin wake envelope. Identified here is a physical source for some of these image artifacts- namely the divergent part of the Kelvin wake system. The SAR images a part of this wave subsystem via a Bragg scattering mechanism, resulting in a narrow vee artifact when the ship and SAR platform directions sufficiently agree. This artifact disappears when surface roughness is present, as with a local breeze, in conformance with observation. The predicted "down-axis" attenuation rate agrees with actual measurements from SAR imagery. A brief comparison of actual SAR images with a those of a simplified SAR simulation is discussed.